HDAC1 and HDAC2 Control the Specification of Neural Crest
Cells into Peripheral Glia
Claire Jacob,
1,4* Pirmin Lo¨tscher,
1* Stefanie Engler,
1,4Arianna Baggiolini,
3Sandra Varum Tavares,
3Vale´rie Bru¨gger,
4Nessy John,
3Stine Bu¨chmann-Møller,
3Paige L. Snider,
5Simon J. Conway,
5Teppei Yamaguchi,
2Patrick Matthias,
2Lukas Sommer,
3Ned Mantei,
1and Ueli Suter
11Institute of Molecular Health Sciences, Department of Biology, ETH Zurich, CH-8093 Zurich, Switzerland,2FMI for Biomedical Research, Novartis Research Foundation, CH-4058 Basel, Switzerland,3Institute of Anatomy, University of Zurich, CH-8057 Zurich, Switzerland,4Institute of Zoology, Department of Biology, University of Fribourg, CH-1700 Fribourg, Switzerland, and5Developmental Biology and Neonatal Medicine Program, H.B. Wells Center for Pediatric Research, Indiana University School of Medicine, Indianapolis, Indiana 46202
Schwann cells, the myelinating glia of the peripheral nervous system (PNS), originate from multipotent neural crest cells that also give
rise to other cells, including neurons, melanocytes, chondrocytes, and smooth muscle cells. The transcription factor Sox10 is required for
peripheral glia specification. However, all neural crest cells express Sox10 and the mechanisms directing neural crest cells into a specific
lineage are poorly understood. We show here that histone deacetylases 1 and 2 (HDAC1/2) are essential for the specification of neural crest
cells into Schwann cell precursors and satellite glia, which express the early determinants of their lineage myelin protein zero (P0) and/or
fatty acid binding protein 7 (Fabp7). In neural crest cells, HDAC1/2 induced expression of the transcription factor Pax3 by binding and
activating the Pax3 promoter. In turn, Pax3 was required to maintain high Sox10 levels and to trigger expression of Fabp7. In addition,
HDAC1/2 were bound to the P0 promoter and activated P0 transcription. Consistently, in vivo genetic deletion of HDAC1/2 in mouse
neural crest cells led to strongly decreased Sox10 expression, no detectable Pax3, virtually no satellite glia, and no Schwann cell precursors
in dorsal root ganglia and peripheral nerves. Similarly, in vivo ablation of Pax3 in the mouse neural crest resulted in strongly reduced
expression of Sox10 and Fabp7. Therefore, by controlling the expression of Pax3 and the concerted action of Pax3 and Sox10 on their
target genes, HDAC1/2 direct the specification of neural crest cells into peripheral glia.
Key words: histone deacetylases; neural crest cells; peripheral glia specification; transcriptional control
Introduction
Specification of neural crest cells into the Schwann cell lineage is
instructed by neuregulin 1 (Shah et al., 1994) and requires the
transcription factor Sox10, which induces the expression of the
neuregulin receptor ErbB3 (Britsch et al., 2001;
Prasad et al.,
2011) and of the early determinants of Schwann cell precursors
fatty acid binding protein 7 (Fabp7) and myelin protein zero (P0)
(Kurtz et al., 1994;
Hagedorn et al., 1999;
Jessen and Mirsky,
2005). Sox10 is expressed in neural crest cells and remains
ex-pressed in peripheral glia, whereas it is downregulated in other
neural crest-derived cells (Kuhlbrodt et al., 1998). Pax3 is another
key transcription factor for neural crest cells and several of their
derived lineages (Auerbach, 1954). In the absence of Pax3, dorsal
root ganglia (DRG) are smaller (Auerbach, 1954;
Olaopa et al.,
2011) and Schwann cells are strongly reduced in number or
ab-sent (Franz, 1990). In the mouse Schwann cell lineage, Pax3 is
expressed until embryonic day 13.5 (E13.5) and is then turned off
and reexpressed at later stages, where it participates in the control
of the Schwann cell cycle (Kioussi et al., 1995;
Doddrell et al.,
2012).
Although Sox10 is clearly critical for peripheral glia
specifica-tion, the mechanisms controlling Sox10 expression and activity
in neural crest cells during this process remain unclear. As for
Pax3, canonical Wnt signaling has been shown to activate the
Pax3 promoter through Cdx transcription factors in Neuro2a
cells (Sanchez-Ferras et al., 2012) to induce sensory neurogenesis
and melanocyte formation, but not specification of peripheral
glia (Hari et al., 2012). Therefore, the mechanisms regulating
Pax3 expression in neural crest cells and Pax3-dependent control
of peripheral glia specification are not known.
We investigated the potential involvement of histone
deacety-lases (HDACs) in this process. HDACs are transcriptional
regu-lators that deacetylate histones to modify chromatin architecture
(de Ruijter et al., 2003;
Michan et al., 2007) or deacetylate
tran-Author contributions: C.J., P.L., L.S., N.M., and U.S. designed research; C.J., P.L., S.E., A.B., S.V.T., V.B., N.J.,S.B.-M., P.L.S., and S.J.C. performed research; C.J., P.L., S.E., S.J.C., T.Y., and P.M. contributed unpublished reagents/ analytic tools; C.J., P.L., S.E., V.B., and L.S. analyzed data; C.J. and U.S. wrote the paper.
This work was supported by the Swiss National Science Foundation and National Center for Competence in Research Neural Plasticity and Repair (C.J., U.S.), the Novartis Research Foundation (P.M.), and the National Insti-tutes of Health (Grant HL60714 to S.J.C.). We thank B. Weinberg, E. Seto, S. Schreiber, W.J. Pavan, G. Marfany, J. Epstein, E. Campeau, G. Lemke, I. Skerjanc, and M.D. Goulding for providing constructs; A. McMahon and P. Soriano for transgenic animals; and the members of the Suter laboratory for useful discussions.
*C.J. and P.L. contributed equally to this work.
Correspondence should be addressed to either of the following: Dr. Claire Jacob, Department of Biology, Institute of Zoology, University of Fribourg, Chemin du Musée 10, 1700 Fribourg, Switzerland, E-mail: [email protected]; or Dr. Ueli Suter, Department of Biology, Institute of Molecular Health Sciences, ETH Ho¨nggerberg, Schafmattstrasse 22, 8093 Zurich, Switzerland, E-mail: [email protected].
Published in 7KH-RXUQDORI1HXURVFLHQFH ± which should be cited to refer to this work.
scription factors to modulate their activity (Glozak et al., 2005;
Yao et al., 2011). Histone deacetylation was long thought to only
result in gene silencing by local compaction of the chromatin
structure. However, HDACs are also needed to keep genes in an
active state (Wang et al., 2009) by acting in concert with histone
acetyltransferases (HATs). Therefore, the level of histone
acety-lation and the presence of HDACs at a gene locus do not always
reflect the activity status of a gene. Here, we elucidate HDAC
functions in conjunction with their binding partners that allow
recruitment to their critical target genes to modulate
transcrip-tional activity. We demonstrate that the highly homologous
HDAC1 and HDAC2 are essential for the specification of neural
crest cells into Schwann cell precursors and satellite glia.
HDAC1/2 induced Pax3 expression in neural crest cells. In turn,
Pax3 maintained high Sox10 levels during specification.
HDAC1/2 also activated P0 expression in neural crest cells. These
mechanisms were necessary for expression of early determinants
of Schwann cells and satellite glia and therefore for their
specification.
Materials and Methods
Plasmids. Sox10 shRNA-1 (MISSION shRNA clone reference
#TRCN0000239154; Sigma) and Sox10 shRNA-2 (MISSION shRNA clone reference #TRCN0000244290; Sigma), Mouse Sox10 MCS1C (plasmid 20240; Addgene;Antonellis et al., 2008), mouse Sox10 MCS4 (kind gift from A. McCallion, Johns Hopkins University School of Medicine, Baltimore, MD), human Fabp7 promoter (kind gift from G. Marfany, Universitat de Barcelona, Barcelona, Spain), rat P0 promoter (kind gift from G. Lemke, The Salk Institute, La Jolla, CA), mouse Pax3 promoter (kind gift from I. Sker-janc, University of Ottawa, Ontario, Canada), pBJ5 HDAC1-overexpression construct (kind gift from S. Schreiber, Broad Institute, Cambridge, MA), pME1 8 S HDAC2-overexpression construct (kind gift from E. Seto, Lee Moffitt Cancer Center and Research Institute, Tampa, FL), Sox10-overexpression construct (kind gift from M. Wegner, Universität Erlangen-Nürnberg, Erlangen, Germany), and Pax3 overexpression construct (Addgene plasmid 27319,Epstein et al., 1995). Constructs for in situ hybrid-ization were as follows: Hdac1 (EST clone, IMAGE ID: 4976514), Hdac2 (mouse full-length cDNA was cloned into the pCMV-Sport6 expression vector), Pax3 (kind gift from M. D. Goulding, The Salk Institute for Biolog-ical Studies, La Jolla, CA), P0 (kind gift from G. Lemke, The Salk Institute, La Jolla, CA and R. Axel, Columbia University, New York, NY), and Fabp7 (EST clone, IMAGE ID: 5700428).
Generation of lentiviruses and transduction of JoMa1 cells. We
previ-ously described HDAC1 shRNA and control shRNA lentiviral constructs (Jacob et al., 2011). We designed a mouse HDAC2-specific shRNA and a Pax3-specific shRNA using Dharmacon siDESIGN Center and 2-shRNA Oligo Designer (http://www.unc.edu/⬃cail/biotool/2shRNA.html).
Targeting sequences were as follows: H2sh, GACCGTCTCATTC-CATAAA; Pax3sh, GGAAGGAAGCAGAAGAAAG. We cloned the cor-responding shRNA oligonucleotides (synthesized by Microsynth) into pLentiLox 3.7 lentiviral plasmid (ATCC) using either HpaI/XhoI or HpaI/NotI restriction sites. For Pax3sh, we replaced GFP with DsRed2, as described previously (Jacob et al., 2011).
To generate the HDAC2-overexpressing lentiviral construct and a GFP control construct, we first replaced the U6 promoter of pSiCoR-GFP with the EF1␣ promoter. pSiCoR-GFP was digested with XbaI, blunted, and redigested with NotI. pEF-ENTR A (plasmid 17427, Add-gene;Campeau et al., 2009) was digested with NcoI, blunted, and redi-gested with NotI to excise the EF1␣ promoter that was religated into digested pSiCoR-GFP. We then digested pSiCoR-GFP-EF1␣ with BamHI and NotI and ligated in mouse HDAC2 cDNA generated by PCR using the HDAC2-overexpressing construct as a template. We called this construct pSiCoR-GFP-EF1␣-HDAC2. PCR primers for HDAC2 were as follows: forward 5⬘-ccGGATCCATGGCGTACAGTCAAGGAGGC-3⬘, reverse 5⬘-ccGCGGCCGCTCAAGGGTTGCTGAGTTGTTC-3⬘. The pSiCoR-GFP-EF1␣ construct was used as a control construct in experi-ments using pSiCoR-GFP-EF1␣-HDAC2. To produce lentiviral
parti-cles, we followed the recommendations of the ViraPower Lentiviral Expression Systems Manual (Invitrogen). JoMa1 cells were transduced at a multiplicity of infection of 2–5 and washed 48 h after transduction.
Cell culture. The JoMa1 neural crest cell-derived cell line was cultured
and induced to differentiate into glia as described previously (Maurer et al., 2007) with modifications. JoMa1 cells were kept on plates that had been coated with poly-D-lysine (1 mg/ml; Sigma-Aldrich) for 5 min and
fibronectin (1 mg/ml) for 1 min. Cells were maintained in DMEM/ Ham’s F12 medium (1:1; Invitrogen) supplemented with: 1% N2 (Invit-rogen), 2% B27 (Invit(Invit-rogen), 10 ng/ml EGF (Peprotech), 4 ng/ml FGF (Peprotech), 100 U/ml penicillin-streptomycin (Invitrogen), and 10% chick embryo extract (both homemade and purchased from Seralab). Proliferation and undifferentiated state were maintained by adding 200 nM4-hydroxytamoxifen (Sigma). For differentiation into glial cells,
ta-moxifen was withdrawn for 4 h before the addition of 5Mforskolin
(Sigma) and 50 ng/ml heregulin-1 (Peprotech) and the cells were main-tained in this medium for the indicated times mentioned in the figure legends (medium was replaced every second day).
Neural crest explants were obtained from E8.75 Naval Medical Re-search Institute mouse embryos. Neural tubes were isolated and plated to enable neural crest cells to migrate out for 12 h, as described previously (Greenwood et al., 1999). Neural tubes were then removed and cells were switched to complete medium (Stemple and Anderson, 1992) containing 50 ng/ml heregulin-1 and 5 Mforskolin to induce
differ-entiation into glia. At this time, 0.6Mmocetinostat (an
HDAC1/2-specific inhibitor; Selleckchem) or its vehicle was also added and cells were incubated for 56 h. This medium was replaced every day.
Generation of conditional knock-out mice. To ablate HDAC1 and HDAC2
in neural crest cells, mice homozygous for floxed Hdac1 and floxed Hdac2 alleles (RRID:MGI:4440556 and RRID:MGI:4440559, respectively; Yama-guchi et al., 2010) were crossed with mice expressing the Cre recombinase under control of the Wnt1 promoter (Wnt1-Cre; RRID:MGI:2386570; Dan-ielian et al., 1998). For in vivo fate mapping, Wnt1-Cre;Hdac1 floxed/wild-type;Hdac2 floxed/wild-type mice were crossed with Hdac1 floxed/floxed;
Hdac2 floxed/floxed;R26R-Lacz⫹/⫹mice in which the Lacz reporter gene has been inserted into the ROSA26 locus preceded by a floxed stop sequence (RRID:MGI:1861932;Soriano, 1999). Genotypes were determined by PCR on genomic DNA. Animal use was approved by the veterinary office of the Canton of Zu¨rich, Switzerland.
Whole-mount X-Gal staining. Embryos harvested at different time
points of gestation were fixed for 5–10 min in 4% PFA and 0.2% glutar-aldehyde in PBS. After extensive washes in PBS, embryos were stained in X-Gal staining solution containing the following: 5 mMK3Fe(CN)6, 5 mM
K4Fe(CN)6, 2 mMMgCl2, 0.1% sodium deoxycholate, 0.02% NP-40, and
1 mg/ml X-Gal in PBS for 1 to several hours at 37°C.
Immunofluorescence and in situ hybridization. Embryos were fixed for
1– 4 h in 4% PFA, cryopreserved overnight in 30% sucrose, and embed-ded in optimal cutting temperature compound (Tissue-Tek). Cryosec-tions (10 or 20m) were submitted to antigen retrieval in citrate buffer at 65°C for 2 h before the blocking step. Neural crest explants were fixed for 15 min at room temperature (RT) in 4% PFA. Cryosections and neural crest explants were incubated in blocking buffer (10% goat serum, 0.3% Triton X-100, 0.2% Tween 20 in PBS or 1% BSA, 0.3% Triton X-100, 0.2% Tween 20 in PBS for primary goat antibodies) for 30 min, and then with primary antibodies overnight at 4°C. Secondary antibodies were applied for 1 h and DAPI for 5 min at RT.
The primary antibodies used were as follows: HDAC1 (1:1000, PA1-860, rabbit polyclonal; Affinity Bio Reagents, RRID:AB_2118091), HDAC2 (1:200, H2663, mouse monoclonal; Sigma Aldrich, RRID: AB_261977), cleaved Caspase 3 (1:500, 9661, rabbit polyclonal; Cell Sig-naling Technology), Sox10 (1:200, N-20, sc-17342, goat polyclonal; Santa Cruz Biotechnology, RRID:AB_2195374; or 1:200, SI058C01, rab-bit polyclonal; DCS, RRID:AB_2313583), Pax3 (1:100, mouse monoclo-nal; Developmental Studies Hybridoma Bank, RRID:AB_528426), NeuroD1 (1:200, sc-1084, goat polyclonal; Santa Cruz Biotechnology, RRID:AB_630922), and Islet1 (1:200, 39.4D5, mouse monoclonal; De-velopmental Studies Hybridoma Bank, RRID:AB_528173).
The secondary antibodies used were as follows: donkey anti-mouse Alexa Fluor 488 (1:500; Invitrogen, RRID:AB_141607), donkey
rabbit Cy5 (1:200; Jackson ImmunoResearch, RRID:AB_2307444), don-key anti-rabbit Alexa Fluor 488 (1:500; Jackson ImmunoResearch, RRID: AB_2313584), and donkey anti-goat Cy3 (1:500; Jackson ImmunoResearch, RRID:AB_2307351).
Cells or sections were observed using a fluorescence microscope (Axioplan2 Imaging; Carl Zeiss) with 20⫻/0.50 numerical aperture (NA), 40⫻/0.75 NA, or 63⫻/1.25 NA (oil-immersion) plan Neofluar objectives. Images were digitized with a PowerShot G5 camera (Canon) and acquired with AxioVision 4.5 software (Carl Zeiss). For confocal analyses, cells were observed using a Leica DMIRE2 inverse microscope and a Leica SP2 AOBS point laser scanning confocal. Optical sections were collected at 2.0m intervals using Leica LCS software. Images were assembled and the brightness and contrast were adjusted using Adobe Photoshop CS5 12.0.4. Where indicated in the figure leg-ends, z-series projections were generated using ImageJ 1.46r software (RRID:nif-0000-30467). Quantification analyses were done with Cell-Profiler 2.0 software (Broad Institute, RRID:nif-0000-00280).
In situ hybridization with digoxigenin-labeled riboprobes was
per-formed on cryosections as described previously (Paratore et al., 1999). Hybridization signals were visualized with NBT/BCIP (Roche Diagnos-tics). Antisense riboprobes were labeled with digoxigenin according to the manufacturer’s instructions (Roche Diagnostics).
Western blot and immunoprecipitation. Twenty-four hour
differenti-ated for immunoprecipitations, IPs, or 4 d differentidifferenti-ated (for Western blot) or undifferentiated JoMa1 cells were lysed. IPs were performed as described previously (Jacob et al., 2011). Immunoprecipitating antibod-ies were cross-linked to Protein A/G PLUS Agarose beads (Santa Cruz Biotechnology) according to the online protocol of New England Bio-Labs. Proteins were eluted in 0.1Mglycine, pH 2.5, and 1MTris, pH 8.0, was added for neutralization. Lysates and IPs were submitted to SDS-PAGE and analyzed by Western blotting.
The primary antibodies used were as follows: HDAC1 (1:2000, PA1-860, rabbit polyclonal; Affinity BioReagents), HDAC2 (1:2000, H2663, mouse monoclonal; Sigma Aldrich), Sox10 (N-20, sc-17342, goat poly-clonal, Santa Cruz Biotechnology, or mouse monopoly-clonal, R&D Systems, RRID:AB_2195180, both at 1:500), Pax3 (1:1000, mouse monoclonal; Developmental Studies Hybridoma Bank), Fabp7 (1:500, rabbit poly-clonal; Abcam, RRID:AB_880078), GAPDH (1:5000, mouse monoclo-nal; Hytest, RRID:AB_1616722). All secondary antibodies were from
Jackson ImmunoResearch: goat anti-mouse HRP (RRID:AB_2313585), goat anti-rabbit HRP (RRID:AB_2313586), and donkey anti-goat HRP (RRID:AB_2313587). Two micrograms of antibody were used per IP.
Luciferase gene reporter assay. JoMa1 cells were transfected in
undiffer-entiated or differentiating conditions. For HDAC1/2 inhibition, cells were first incubated with 0.6MMocetinostat (Selleckchem) for 48 h in undifferentiated conditions before transfection and induction of differ-entiation. Luciferase activity was first normalized to a luciferase-empty control (transfected with the overexpressing construct at the same ratio Fugene 6:DNA), and then normalized to-galactosidase activity (assay according to the instructions of Promega). Results are expressed as per-centage of the control.
For transfection, Fugene 6 (Roche) was used at a 3:1 ratio Fugene 6:DNA, according to the manufacturer’s recommendations with the following modification: the DNA was incubated in Optimem for 5 min at room temperature before addition of Fugene 6. The mix Fu-gene 6:DNA was then incubated for 25 min at room temperature before being added to cells. Cells were lysed in 110l of reporter lysis buffer (Promega) and assayed for luciferase activity. Forty microliters of lysate were subjected to two consecutive injections of 25l of luciferase substrate and values were recorded after an integration time of 30 s. Efficiency of transfection was evaluated by measuring -galactosidase activity of a cotransfected lacZ construct (Promega). Beta-galactosidase activity assay was as follows: 50l of assay buffer 2⫻ was added to 50 l of lysate and -galactosidase activity was recorded after a 15–20 min incubation time at 37°C. Endogenous -galactosidase activity (as determined from cells transfected only with overexpressing or shRNA constructs at the same ratio Fugene 6:DNA) was subtracted.
We have verified that no significant change of luciferase activity of the empty luciferase construct pGL3 (backbone construct; Promega) occurred when overexpressing or shRNA constructs were transfected. Cells were cotransfected with 0.8g of luciferase construct (Sox10 MCS4, Sox10 MCS1C, P0 promoter, Pax3 promoter, Fabp7 promot-er); 0.8g of GFP-, HDAC1-, or HDAC2-overexpressing construct; or between 0.05 and 0.8g of Sox10- or Pax3-overexpressing con-struct, 0.8g of shRNA construct (control shRNA, HDAC1 shRNA, HDAC2 shRNA, Pax3 shRNA, Sox10 shRNAs), and 300 ng of lacZ construct.
Figure 1. HDAC1/2 are robustly expressed in DRG and upregulated in JoMa1 neural crest cells during specification into glia. A, In situ hybridization of Hdac1 and Hdac2 on transverse neural tube
sections of E10.5 and E11.5 wild-type mouse embryos (representative pictures of three embryos analyzed). Arrows indicate DRG. Scale bar, 100m. Western blot pictures of Sox10 (B), Fabp7 (C), HDAC1 (D), and HDAC2 (E) and quantification normalized to GAPDH in 4 d differentiated (Diff.) versus undifferentiated (Undiff. ⫽ 1) JoMa1 cells. Dividing lines indicate that samples have been loaded on the same gel, but not on consecutive lanes. p-values (two-tailed paired Student’s t test unless stated otherwise in the figure): 0.0418 (B), 0.0157 (C), 0.0026 (D), 0.011 (E). Results are from at least three independent experiments. Error bars indicate SEM.
Quantitative real-time PCR. Isolation of RNA was performed using
TRIzol reagent (Invitrogen) according to the manufacturer’s recommen-dations. cDNA was produced using SuperScript III Reverse Transcriptase (Invitrogen). Quantitative real-time PCR (qRT-PCR) analyses were per-formed on an ABI 7000 Sequence Detection System (Applied Biosys-tems) using 2⫻ SYBR Green PCR Mastermix (Applied Biosystems) according to the manufacturer’s recommendations.
Primer sequences were as follows: for P0, forward 5 ⬘-CCCTGGCCAT-TGTGGTTTAC-3⬘, reverse 5⬘-CCATTCACTGGACCAGAAGGAG-3⬘; for GAPDH, forward 5⬘-CGTCCCGTAGACAAAATGGT-3⬘, reverse 5⬘-TTGATGGCAACAATCTCCAC-3⬘. The amplification program was as follows: 5 min at 96°C; 40 steps of 30 s at 96°C, 30 s at 60°C, 30 s at 72°C, and 5 min at 72°C.
ChIP. Experiments with JoMa1 neural crest cells were performed as
described previously (Jacob et al., 2011). Two micrograms of antibodies were used per IP.
The primer sequences for SYBR Green qRT-PCR were as follows: Sox10 MCS4, forward 5⬘-AACGAGGGATGCAAGGAAGG-3⬘, reverse 5⬘-GGGAACAATGTCAACGTCGC-3⬘; Pax3 promoter, forward 5⬘-TCCACAGGTGAAAAGCAGGG-3⬘, reverse 5⬘-GGTGTCCCTGTC-CCTCTACA-3⬘; P0 promoter, forward 5⬘-CCCCCTACCCTAGGTT GGAA-3⬘, reverse 5⬘-CCTTTGCCATGGCCTTTTGG-3⬘; Fabp7 promoter, forward 5⬘-AGTTTTCTGTTCCCAGGCCC-3⬘, reverse 5⬘-TGGGTTAGCGGAGTAGGTCA-3⬘; Sox10 MCS1C, forward 5⬘-GGCAATGGTTTAGGGTCCCA-3⬘, reverse 5⬘-TTGGCATCTGGTG TGAGCAT-3⬘; control sequence (localized at ⫺2.3kb from transcription start site of P0 gene), forward 5 ⬘-ATCCAAAGCCAAAAAGAAAAGTGC-3⬘, reverse 5⬘-GTACGGGTACTTACAACAAATGGC-3⬘.
The antibodies used were as follows: mouse anti-HDAC2 (H2663; Sigma Aldrich), rabbit anti-HDAC1 (PA1– 860; Affinity BioReagents), goat anti-Sox10 (N-20, sc-17342; Santa Cruz Biotechnology), mouse anti-Pax3 (Developmental Studies Hybridoma Bank), mouse anti-HH3 (positive control; Millipore, RRID:AB_309763), or rabbit anti-mouse IgG (negative control; Sigma, RRID:AB_260634).
Results
HDAC1 and HDAC2 are robustly expressed during
specification of neural crest cells into peripheral glia
We found that HDAC1 and HDAC2 were widely expressed in the
neural tube and surrounding tissue at E10.5 in developing mouse
embryos (Fig. 1
A). At E11.5, however, HDAC1/2 expression was
concentrated in the neural tube and in DRG (Fig. 1
A) and
re-mained highly expressed in these structures and in peripheral
nerves at E12.5 and E14.5 (data not shown). Specification of
neu-ral crest cells into peripheneu-ral glia occurs at E11 (Jessen and
Mir-sky, 2005). Because HDAC1/2 expression was maintained during
this process in DRG, we investigated the potential involvement of
HDAC1/2 in the transcriptional control of peripheral glia
speci-fication. We initially used the neural crest cell-derived line
JoMa1, which differentiates into glia upon treatment with the
neuregulin 1 isoform heregulin-beta1 (Maurer et al., 2007). Within
4 d, heregulin-beta1 upregulated Sox10 (Fig. 1
B) and the early glial
marker Fabp7 (Fig. 1
C) and led to a mild but significant
upregula-tion of HDAC1 (Fig. 1
D) and HDAC2 (
Fig. 1
E) in these cells.
Figure 2. HDAC1/2 activity is required to maintain high Sox10 levels and for Sox10-dependent activation of Sox10, Fabp7 and P0 gene regulatory regions during specification into glia. A,
Immunofluorescence of Sox10 (red; DAPI⫽nucleiinblue)onmouseneuralcrestexplantsindifferentiatingconditionswithorwithouttheHDAC1/2inhibitormocetinostat(HDAC1/2i;0.6M), and
quantification of Sox10 (200 cells per explant, 3 explants per condition, n ⫽ 3). Relative luciferase activity of Sox10 MCS4 (B,C), Sox10 MCS1C (⫽ Sox10 promoter region, B), Fabp7 promoter (D), and P0 promoter (E)-driven luciferase reporter constructs (-luc) induced by Sox10 overexpression compared with GFP (⫽ 1) with or without HDAC1/2i (C–E) 24 h after transfection and induction of differentiation in postnatal primary rat Schwann cells (B) or in JoMa1 cells (B–E). p-values from two-tailed unpaired (A) or paired (B–E) Student’s t test: 0.0063 (A); *0.0193, ***6.91462E-06 (B); **0.0029, ***0.0001,ⴙ0.026 (C); **1.00025E-05, ***0.0001, ⴙ0.017 (D); and **0.0041, ⴙⴙ0.028 (E). Results are from at least three independent experiments. Error bars indicate SEM.
Interestingly, a second HDAC2 band of slightly higher
molec-ular weight appeared in glia-differentiated JoMa1 cells,
indi-cating posttranslational modification of HDAC2 during this
process. Together, these data confirmed that the JoMa1 neural
crest cell line can be specified into glia upon heregulin-beta1
treatment and was therefore suitable for our study. In
addi-tion, the robust expression of HDAC1/2 concomitant with
Sox10 upregulation prompted us to seek a potential
involve-ment of HDAC1/2 in the regulation of peripheral glia
specification.
HDAC1/2 activity is required to maintain high Sox10
expression and for Sox10-dependent activation of Sox10,
Fabp7, and P0 gene regulatory regions during specification
into glia
Precise regulation of Sox10 expression is critical for specification
of neural crest cells into glia (Paratore et al., 2001). To determine
whether HDAC1/2 activity is involved in the control of Sox10
expression during this process, we chose to use mouse neural
crest explants because the myc-immortalized JoMa1 cell line may
not completely mirror all transcriptional regulations that occur
in normal neural crest. We induced glia specification in mouse
neural crest explants in the presence of the HDAC1/2-specific
inhibitor mocetinostat (0.6
M) and analyzed Sox10 levels. At
this concentration, mocetinostat is a specific inhibitor of
HDAC1/2 (IC50
for HDAC1
⫽ 0.15
M, IC50
for HDAC2
⫽ 0.29
M). It also becomes active on HDAC3 at higher concentrations
(IC50
for HDAC3
⫽ 1.66
M). Inhibition of HDAC1/2 activity
strongly decreased Sox10 expression (Fig. 2
A) without a
signifi-cant increase in cell death (data not shown), indicating that
HDAC1 and/or HDAC2 is necessary to maintain high Sox10
ex-pression during specification of neural crest cells into glia.
To determine the underlying molecular mechanism, we
started from the fact that Sox10 upregulates its expression by
binding and activating its own promoter and 5
⬘ enhancers (
Wer-ner et al., 2007;
Antonellis et al., 2008). Activation of the Sox10
Figure 3. HDAC2 binds to and activates the P0 promoter, whereas HDAC2-dependent activation of Sox10 MCS4 and the Fabp7 promoter is most likely indirect. Relative luciferase activity of P0
promoter-luc (A,D),Sox10MCS4-luc(E),andFabp7promoter-luc(F)byHDAC1(H1)orHDAC2(H2)overexpressionwithorwithoutSox10shRNAs(S10sh;D)orinthepresenceofSox10shRNAsalone (D) compared with GFP (⫽ 1) or GFP/Control shRNA (GFP/Csh ⫽ 1), 24 h (A) or 24 and 60 h (E, F) after transfection and induction of differentiation or 72 h (D) after transfection and 24 h in differentiating conditions.B, Fold increase of P0 mRNA induced by HDAC1, HDAC2, or Sox10 overexpression compared with GFP (⫽ 1) and normalized to GAPDH mRNA 48 h after transfection and induction of differentiation.C,ImmunofluorescenceofSox10(green)andDAPI(blue⫽nuclei)labelingofmouseDRGexplantcultures3daftertransfectionwithcontrolshRNA,Sox10shRNA-1,and Sox10 shRNA-2 constructs.G, ChIP of HDAC1 (H1) and HDAC2 (H2) to P0 promoter (P0 pro) or control sequence (control seq.) compared with negative control (IgG ⫽ neg ⫽ 1). H, Immunopre-cipitation of HDAC2 or Sox10 in undifferentiated or 24 h differentiated JoMa1 cells and Western blot for HDAC2. Quantification of coimmunoprecipitated HDAC2 normalized to Sox10 input. p-values (two-tailed paired Student’s t test, unless stated otherwise): *0.041 and ***0.00046 (A); *0.042, **0.0058, ***8.4726E-05 (B); *0.037, 0.041, 0.03, ***0.00004 (D, p-values compared with the control GFP/Csh),ⴙ0.044,0.031,0.029,0.029(D,p-valuescomparedwithH1/CshforH1/S10shorH2/CshforH2/S10sh);*0.044(E);0.03(F);0.044(G);and***0.0007(H).Resultsarefromatleast three independent experiments. Error bars indicate SEM.
MCS4 enhancer (also called U3) by Sox10 is critical for Sox10
expression in neural crest cells (Wahlbuhl et al., 2012). First, we
used the luciferase gene reporter assay to test the activity of the
Sox10 MCS4 enhancer and of Sox10 MCS1C (Sox10 promoter) in
JoMa1 cells and postnatal primary Schwann cells upon Sox10
overexpression (Fig. 2
B,C). Sox10 MCS4 was strongly activated in
JoMa1 cells, whereas Sox10 MCS1C was not. In contrast, Sox10
MCS4 was not activated in postnatal Schwann cells, whereas
Sox10 MCS1C was (
Fig. 2
B). These findings confirm the findings
of
Werner et al. (2007)
and extend the importance of these two
regulatory regions at different stages of development. Sox10
overexpression also robustly activated the promoters of the early
determinants of Schwann cells, Fabp7 (Fig. 2
D) and P0 (
Fig. 2
E),
in JoMa1 cells.
Exogenous DNA transiently applied in cells is present in
nucleosomal DNA at approximately the same levels as
endoge-nous DNA (Zhang et al., 2002). We therefore tested the effects of
HDAC1/2 inhibition in these validated settings. Inhibition of
HDAC1/2 activity decreased Sox10-induced activation of Sox10
MCS4 (Fig. 2
C), the Fabp7 promoter (
Fig. 2
D), and the P0
pro-moter (Fig. 2
E). Therefore, HDAC1 and/or HDAC2 are needed
for Sox10-dependent activation of these regulatory elements.
P0 is a direct target gene of HDAC2, whereas
HDAC2-dependent activation of Sox10 MCS4 and the Fabp7 promoter
is most likely indirect
In JoMa1 cells cultured in differentiating conditions,
overexpres-sion of HDAC2 and, to a lesser extent, of HDAC1 increased the
activation of the P0 promoter, as detected by luciferase gene
re-porter assay 24 h after transfection (Fig. 3
A), and increased the
transcript levels of P0 48 h after transfection (Fig. 3
B). To bind
DNA, HDACs need to interact with a DNA-binding protein.
Be-cause Sox10 binds and activates the P0 promoter (Peirano and
Wegner, 2000;
Peirano et al., 2000;
Fig. 2
E) and increases P0
transcript levels (Fig. 3
B), we tested whether HDAC1/2-induced
activation of the P0 promoter requires Sox10 by downregulating
Sox10 with specific shRNAs (Fig. 3
C). Indeed, this activation was
suppressed when Sox10 was downregulated (Fig. 3
D), indicating
that HDAC1/2-mediated activation of the P0 promoter is Sox10
dependent.
HDAC2 activated Sox10 MCS4 and the Fabp7 promoter 60 h
after transfection (Fig. 3
E,F ), but not after 24 h, suggesting an
indirect activation mechanism. In contrast, the P0 promoter is
likely to be activated in a direct manner. Indeed, using ChIP, we
found HDAC2 bound to the P0 promoter 24 h after induction of
Figure4. HDAC2controlsPax3expressioninneuralcrestcells.A,WesternblotofPax3andquantificationnormalizedtoGAPDHin4ddifferentiated(Diff.)comparedwithundifferentiated(Undiff.
⫽1)JoMa1cells.B,ImmunofluorescenceofPax3(green;DAPI⫽nucleiinblue)inneuralcrestexplantsindifferentiatingconditionswithorwithouttheHDAC1/2inhibitormocetinostat(HDAC1/2i, 0.6M) and quantification of Pax3 (⬃200cellsperexplant,3explantspercondition,n⫽3).RelativeluciferaseactivityofPax3promoter-lucuponHDAC1(H1),HDAC2(H2)orSox10overexpression
(C)withorwithoutSox10shRNAs(S10sh)orinthepresenceofSox10shRNAsalone(F)comparedwithGFP(⫽1)orGFP/controlshRNA(GFP/Csh⫽1)24or72h(shRNA)aftertransfectionand24h of induction of differentiation or in the presence of H1 shRNA (sh) or H2sh compared with Csh (⫽1)(D)72haftertransfectionand24hindifferentiatingconditions.E,WesternblotofHDAC2,Pax3, and GAPDH in JoMa1 cells expressing endogenous HDAC2 (C, control) or overexpressing HDAC2 (H2o). A lentiviral vector was used to overexpress HDAC2. Control cells were transduced with a lentiviral vector expressing GFP. Cells were collected 72 h after transduction and induction of differentiation. Dividing lines indicate that samples have been loaded on the same gel, but not on consecutive lanes.G, ChIP showing relative binding of anti-HDAC1 (H1) and anti-HDAC2 (H2) antibodies to Pax3 promoter (Pax3 pro) or control sequence (control seq.) compared with negative control (IgG⫽neg⫽1).H,ChIPshowingrelativebindingofanti-Sox10antibodytoSox10MCS4,Sox10MCS1C,Pax3promoter(pro),Fabp7pro,orcontrolsequenceregion(controlseq.)compared with negative control (IgG⫽ neg ⫽ 1), 24 h after induction of differentiation. p-values from Student’s t test (two-tailed paired in A, C, D, E–H or unpaired in B) unless stated otherwise: 0.044 (A); 0.025 (B); *0.041, **0.0016 (C); 0.0003 (D); 0.04 (E); *0.047, 0.031, 0.048, **0.0032 (F, p-values compared with control GFP/Csh), ⴙ0.026, 0.04, 0.028, 0.027 (F, p-values compared with H1/Csh for H1/S10sh or to H2/Csh for H2/S10sh); *0.027 and **0.0019 (G); 0.033 (one-tailed) and 0.0125 (two-tailed; H). Results are from at least three independent experiments. Error bars indicate SEM.
differentiation in JoMa1 cells (Fig. 3
G).
We previously showed that HDAC1/2
in-teract with Sox10 upon induction of
dif-ferentiation in postnatal Schwann cells to
activate the Sox10 promoter and the
Krox20 myelinating Schwann cell element
(Krox20 MSE;
Jacob et al., 2011). In
JoMa1 cells also, Sox10
coimmunopre-cipitated with HDAC2 and this
interac-tion was significantly increased 24 h after
induction of specification into peripheral
glia (Fig. 3
H ). Together, these data
indi-cate that HDAC2 interaction with Sox10
in neural crest cells increases upon
induc-tion of specificainduc-tion into glia to bind and
activate the P0 promoter.
Pax3 is a direct target gene of HDAC2
in neural crest cells
To understand how HDAC2 controls the
activation of Sox10 MSC4 and the Fabp7
promoter, we looked for an intermediate
HDAC target gene. In JoMa1 cells, Pax3
was upregulated upon differentiation into
glia (Fig. 4
A), which is consistent with
functional roles in this process. To test
whether HDAC1/2 are involved in
con-trolling Pax3 expression, we used mouse
neural crest explants cultured in
glia-differentiation conditions. We found that Pax3 expression was
decreased by HDAC1/2 inhibition (Fig. 4
B), indicating that
HDAC1 and/or HDAC2 is indeed involved in controlling Pax3
expression.
Overexpression of HDAC2 or Sox10 in JoMa1 cells rapidly (24
h after transfection) activated the Pax3 promoter (Fig. 4
C),
whereas downregulation of HDAC2 by shRNA resulted in
de-creased activity (Fig. 4
D). Consistently, HDAC2 overexpression
resulted in increased Pax3 expression levels (Fig. 4
E). Because
HDAC2 interacted with Sox10 in JoMa1 cells (Fig. 3
H ) and both
HDAC2 and Sox10 overexpression activated the Pax3 promoter
(Fig. 4
C), we tested whether Sox10 is involved in
HDAC2-mediated activation of the Pax3 promoter. HDAC2-induced
ac-tivation of the Pax3 promoter was abolished when Sox10 was
downregulated by shRNAs (Fig. 4
F ), showing that this activation
is Sox10 dependent. Furthermore, we found that HDAC1 and
HDAC2 were highly enriched at the Pax3 promoter (Fig. 4
G).
Sox10 was also bound to the Pax3 promoter, but not to Sox10
MCS1C or the Fabp7 promoter (Fig. 4
H ). Consistent with
previ-ous studies (Werner et al., 2007;
Antonellis et al., 2008;
Wahlbuhl
et al., 2012), Sox10 was bound to Sox10 MSC4 (Fig. 4
H ). These
data strongly suggest that HDAC1/2 are recruited by Sox10 to the
Pax3 promoter. We conclude that Sox10 and HDAC2 control
Pax3 expression in neural crest cells directly by binding and
acti-vating the Pax3 promoter.
Pax3 is necessary for HDAC2-dependent activation of Sox10
MCS4 and for expression of early determinants of peripheral
glia during specification
We next hypothesized that Pax3 could be the critical HDAC2
intermediate target gene required for the maintenance of high
Sox10 expression during specification of neural crest cells into
glia. To test this hypothesis, we generated a lentiviral vector that
carries a specific Pax3 shRNA to downregulate Pax3 in neural
crest cells. Pax3 downregulation in JoMa1 cells resulted in
signif-icantly reduced Sox10 and Fabp7 expression (Fig. 5
A), indicating
that Pax3 is necessary for Sox10 and Fabp7 expression in neural
crest cells.
Consistent with previous data (Wahlbuhl et al., 2012), mild
Pax3 overexpression activated Sox10 MCS4 in JoMa1 cells,
whereas high overexpression did not (Fig. 5
B). In contrast, Sox10
MCS4 activation increased with the amount of overexpressed
Sox10 (Fig. 5
B). Pax3-binding sites on Sox10 MCS4 are adjacent
to Sox10-binding sites (Wahlbuhl et al., 2012). Therefore, high
levels of Pax3 may limit the access of Sox10 to its own binding
sites and prevent Sox10-induced activation of Sox10 MCS4. In
addition, Sox10 and Pax3 can interact (Lang and Epstein, 2003),
so an excess of Pax3 could sequester Sox10 away from its target
genes. Because low levels of Pax3 activated Sox10 MCS4 (Fig. 5
B),
residual Pax3 expression in the presence of shRNA is likely to be
sufficient to maintain Sox10 at moderate levels (Fig. 5
A).
Pax3 downregulation reduced Sox10 MCS4 activation and
prevented HDAC2-induced Sox10 MCS4 activation (Fig. 5
C).
Pax3 also rapidly activated the P0 promoter 24 h after
transfec-tion and inductransfec-tion of differentiatransfec-tion (Fig. 5
D), but not the Fabp7
promoter (data not shown). Pax3 is known to bind Sox10 MCS4
(Wahlbuhl et al., 2012). We confirmed this binding in JoMa1
cells and found Pax3 to be additionally enriched at the P0
pro-moter, but not at the Sox10 MCS1C (Fig. 5
E). These data indicate
that Pax3 is required for HDAC2-dependent maintenance of
high Sox10 levels and expression of peripheral glia proteins
dur-ing specification of neural crest cells into glia.
In vivo, HDAC1/2 are essential in neural crest cells for Pax3
expression, maintenance of high Sox10 levels, and
specification into peripheral glia
Next, we wanted to validate the relevance of HDAC1/2 functions
in peripheral glia specification in vivo. Although HDAC1 and
Figure5. Pax3controlsglialmarkerandregulatorexpressionduringspecification.A,Pax3,Sox10,andFabp7Westernblotsand
quantification normalized to GAPDH in JoMa1 cells 6 d after transduction with control shRNA (Csh⫽ 1) or Pax3 shRNA (Pax3sh) lentiviruses and 4 d in differentiating conditions. Luciferase fold induction of Sox10 MCS4-luc in JoMa1 cells (B) transfected with increasing amounts of Pax3 or Sox10 compared with GFP (⫽ 1) 24 h after transfection in differentiating conditions or expressing endogenous or overexpressed HDAC2 (H2) with or without Pax3sh (C) compared with GFP/Csh (⫽ 1) 96 h after transfection and 48 h in differentiating conditions.D, Luciferase fold induction of P0 promoter-luc by Pax3 overexpression compared with GFP (⫽ 1) 24 h after transfection in differentiating conditions.E,ChIPofPax3toP0promoter(P0pro),Sox10MCS4,Sox10MCS1C,orcontrol sequence (control seq.) compared with negative control (IgG⫽ neg ⫽ 1). p-values (one-tailed or two-tailed paired Student’s t test): *0.03, 0.048, **0.0067 (A, one-tailed); *0.0277 (B, one-tailed); *0.0197, 0.0166, ***4.953E-05, 6.91E-6 (B, two-tailed); *0.0355, **0.0016, ***0.0006,ⴙⴙ0.009(C,two-tailed);0.008(D,two-tailed);**0.0079,***0.0001(E,two-tailed).Resultsare from at least three independent experiments. Error bars indicate SEM.
HDAC2 can have distinct primary functions, they efficiently
compensate for the loss of each other (Jacob et al., 2011).
There-fore, we depleted both HDACs in neural crest cells. To this end,
we crossed mice carrying floxed Hdac1 and floxed Hdac2
(Yama-guchi et al., 2010) and expressing the Cre recombinase under
control of the Wnt1 promoter (Wnt1-Cre;
Danielian et al., 1998).
Although no detectable reduction of HDAC1/2 protein levels was
observed at E9.5, HDAC1/2 were efficiently ablated at E10 from
DRG and peripheral nerves of double HDAC1/HDAC2
homozy-gous knock-out embryos (dKO,
Fig. 6
A). For in vivo fate mapping
of neural crest cells, we included a Cre-dependent LacZ reporter
allele on dKO and control embryos. Although smaller than in
control embryos, DRG were formed in dKO embryos at E10.5
and E11.5, as well as peripheral nerves (Fig. 6
B). These findings
indicate that neural crest migration was not massively affected by
the loss of HDAC1 and HDAC2. Consistent with our cell culture
data, Sox10 protein levels were robustly decreased at E10.5 and
E11 in dKO DRG compared with controls (Fig. 6
C). Although the
number of apoptotic cells in DRG was increased in dKO embryos
at E10.5 and E11, Sox10-positive cells in control DRG (normal
Sox10 levels) or in dKO DRG (low Sox10 levels) were rarely
positive for cleaved Caspase 3 (Fig. 6
D). We conclude that cells
expressing Sox10 at detectable levels in dKO DRG did not
un-dergo increased apoptosis. Pax3 was present in Sox10-expressing
cells of control DRG, but absent in cells of dKO DRG (Fig. 6
C).
We confirmed this absence by in situ hybridization (Fig. 6
E). Also
consistent with our in vitro data, Fabp7 and P0 transcripts,
ex-pressed in DRG and/or peripheral nerves, were not detected in
Figure 6. HDAC1/2 are essential for in vivo specification of peripheral glia. A, Coimmunofluorescence of HDAC1 (blue) and HDAC2 (green) in E9.5 and E10 dKO DRG and peripheral nerves. White
arrows indicate DRG and arrowheads peripheral nerves. Three control and three dKO embryos for each age were stained and representative photographs are shown.B, Whole-mount in vivo fate mapping using the recombination-dependent Rosa26-lacZ reporter allele at E10.5 and E11.5. Spinal cords imaged at the same level in control and dKO embryos are shown. Insets on the right are a magnification of two E11.5 DRG and their peripheral nerve (PN). Arrows show DRG or PN, as indicated in the figure.C, Confocal pictures (z-series projections) of Sox10 (red) and Pax3 (green) coimmunofluorescence and DAPI labeling (⫽nucleiinblue)inE11controlandHDAC1/2dKODRG,andSox10quantificationatE10.5andE11(Sox10-expressingcellsinDRGof3controlsand3dKOs for each developmental stage). Overlay of Sox10 and Pax3 appears yellow, and of Sox10 and DAPI appears pink. Arrows indicate some Sox10-expressing cells, arrowheads some nonspecific auto-fluorescence (white).D, Confocal pictures of Sox10 (red) and cleaved Caspase 3 (cCasp3, green) coimmunofluorescence and DAPI labeling (blue) in E11 control and HDAC1/2 dKO DRG, and quantification of apoptotic cells (cCasp3-positive) in DRG at E10.5 and E11 (3 controls and 3 dKOs for each developmental stage). White arrows show cCasp3-positive cells, pink arrowheads show nonspecific auto-fluorescence (magenta). In situ hybridization of Pax3 (E), Fabp7 (F ), and P0 (G) in E11 controls and dKOs (representative pictures of 3 controls and 3 dKOs). Asterisks indicate DRG, arrows, peripheral nerves. p-value from two-tailed paired (C, E10.5) or unpaired (C, E11; D) Student’s t test: *0.02, 0.037 (C); and *0.019, 0.038 (D). Error bar indicates SEM.
these structures in dKO embryos (Fig.
6
F,G). Similar to dKO embryos, Sox10
and Fabp7 expressions were strongly
re-duced in Wnt1-Cre Pax3 knock-out DRG
(Fig. 7
A, B) and DRG were smaller
(Olaopa et al., 2011
and data not shown).
HDAC1/2 are not essential for overall
sensory neurogenesis or for
differentiation of cardiac neural crest
cells
To determine whether HDAC1/2 are also
required for specification of neural crest
cells into sensory neurons or cardiac
ral crest, we used specific markers of
neu-rons in DRG and of differentiated cardiac
neural crest cells in the heart outflow tract
of dKO and control embryos.
There are two waves of sensory
neuro-genesis, the first occurring at E9.5 and the
second at E10 in mouse embryos (Ma et
al., 1999). Newborn sensory neurons
ex-press the transient marker NeuroD1 and
the pan-sensory neuronal marker Islet1.
Therefore, at E10.5, sensory neurons
gen-erated during the first wave express Islet1,
but do not express NeuroD1 anymore, whereas sensory neurons
of the second wave are NeuroD1 and Islet1 positive. At E10 and
E10.5, we detected both NeuroD1/Islet1-double positive cells and
Islet1-positive/NeuroD1-negative cells in DRG of dKO embryos
and controls (Fig. 8
A). We cannot address whether HDAC1/2 are
necessary for the first wave of sensory neurogenesis because onset
of HDAC1/2 loss in dKO neural crest cells is at E10. However, we
can state that specification of neural crest cells into sensory
neu-rons of the second wave occurs in the absence of HDAC1/2.
Nev-ertheless, the number of sensory neurons was reduced in dKO
DRG compared with controls (data not shown). Islet1-positive
cells were rarely positive for cleaved caspase 3 (cCasp3;
Fig. 8
B) in
DRG of E10.5 dKO. Therefore, specified sensory neurons did not
undergo massive apoptosis. Rather, the reduced number of
sen-sory neurons may be explained by reduced proliferation of DRG
sensory neurons in the absence of HDAC1/2 (data not shown).
In vivo fate mapping revealed that cardiac neural crest cells were
also able to migrate to the branchial arches and the heart outflow
tract and to differentiate in the absence of HDAC1/2, as shown by
X-Gal staining and the expression of the differentiation markers
Sox9 and smooth-muscle actin (SMA) in dKO and control embryos
carrying also the Cre-dependent LacZ allele (Fig. 8
C). However, the
number of cardiac neural crest cells was reduced, possibly due to
increased apoptosis, as detected by cCasp3 stainings in dKO
bran-chial arches and heart outflow tracts compared with controls (Fig.
8
C).
Discussion
Schwann cells, the myelinating cells of the PNS, are essential for the
survival and the functions of peripheral neurons. In addition,
Schwann cells are remarkably plastic, which enables them to foster
axonal regeneration after a lesion. Because of their exceptional
regener-ative properties, Schwann cells are under consideration in
transplan-tation paradigms when large lesions of the PNS have occurred and
for regeneration of lesions in the CNS. To use these cells in
regener-ative medicine, it is critical to know how to control their cell cycle
and to understand how they specify and differentiate, all the more so
because stem cells are often used to generate Schwann cells for
transplantation.
Specification of neural crest cells into Schwann cells and satellite
glia requires the transcription factor Sox10 (Britsch et al., 2001).
However, all neural crest cells express Sox10 and the mechanisms
that drive neural crest cells into the peripheral glia lineage remain
unclear. The aim of this study was to elucidate these mechanisms. By
combining the use of the JoMa1 neural crest cell line, mouse neural
crest explant cultures, and mouse genetics, we demonstrate that
HDAC1/2 are essential transcriptional regulators of neural crest
specification into Schwann cells and satellite glia.
HDACs are chromatin-remodeling enzymes that control DNA
access for the transcriptional machinery by removing acetyl groups
from histone tails. DNA is negatively charged and histones are
pos-itively charged. The addition of acetyl groups to histone tails
neutral-izes their positive charges. This leads to less interaction between
DNA and histones, thereby resulting in a more relaxed chromatin
that facilitates the access for transcription factors. In contrast, when
acetyl groups are removed by HDACs from histone tails, histones
recover their positive charges, which leads to a tightly wound DNA
around histones that results in a more condensed chromatin.
Be-cause histone deacetylation favors chromatin compaction, HDACs
are commonly seen as transcriptional repressors. However, the effect
of HDACs on chromatin compaction is more complex. Indeed,
many lysine residues of histones can be acetylated or deacetylated,
allowing for different degrees of chromatin compaction. It is thus
conceivable that, depending on the degree of chromatin
compac-tion, HDACs can either block DNA access for the transcriptional
machinery and thereby silence gene expression or only limit this
access for selected transcription factors to induce expression of
spe-cific genes or activation of spespe-cific gene loci. Another mechanism by
which HDACs control transcription at the chromatin level has been
demonstrated in a study by
Wang et al. (2009)
showing that HATs
and HDACs act in concert to maintain genes in a transcriptionally
active state. Therefore, HDACs do not always repress transcription,
but can also be necessary for transcriptional activity. Our previous
work shows that HDAC1/2 bind to the Sox10 promoter and the
Figure 7. Sox10 and Fabp7 are strongly reduced in Wnt1-Cre Pax3 knock-out DRG. A, Confocal pictures (z-series projections) of
Pax3 (green) and Sox10 (red) coimmunofluorescence, and DAPI (blue) labeling in E11.5 control and Pax3 KO DRG and Sox10 quantification (at least 100 Sox10-expressing cells quantified in DRG of two E11.5 and one E10.5 controls and two E11.5 and one E10.5 Pax3 KOs). When quantifications of all embryos are pooled, averaged Sox10 intensity per cell is significantly decreased by 70% in Pax3 KOs compared with controls (two-tailed unpaired Student’s t test, p-value ⫽ 0.045). White arrows show some Sox10-positive cells; pink arrowheads, nonspecific auto-fluorescence signal.B, In situ hybridization of Fabp7 in E11.5 control and Pax3 KO embryos (representative photographs of two controls and two Pax3 KOs). Arrows indicate DRG.
Krox20 MSE in postnatal Schwann cells to activate the transcription
of their genes and are therefore essential for the myelination process
(Jacob et al., 2011). In addition to their effect at the chromatin level,
HDACs are also able to deacetylate nonhistone targets, including
several transcription factors, to modulate their activity. HDACs are
thus very potent transcriptional regulators that can act at different
levels.
There are 18 known mammalian HDACs that are subdivided
into four classes based on their structure. HDAC1 and HDAC2
be-long to class I and are almost exclusively found in the nuclear
com-partment. These two HDACs are highly homologous and many
studies have reported that they efficiently compensate for the loss of
each other, even though they can have distinct primary functions
when expressed at normal levels (Jacob et al., 2011). Because
HDAC1/2 were robustly expressed at the time of glia specification in
mouse DRG and were upregulated in JoMa1 neural crest cells upon
induction of specification into glia, we became interested in the
po-tential functions of these two HDACs in the specification of neural
crest cells into peripheral glia. During specification, Sox10
expres-sion is maintained in the glial lineage, whereas it is downregulated in
other neural-crest-derived lineages. We found that HDAC1/2
deacetylase activity is required to maintain high Sox10 levels and for
expression of Pax3 in neural crest explants during specification into
glia. Pax3 is also a critical transcription factor for neural crest cells
and neural-crest-derived structures. Consistent with our data using
a specific HDAC1/2 inhibitor in neural crest explants, in vivo genetic
ablation of HDAC1/2 in neural crest cells resulted in strongly
re-duced Sox10 levels, no detectable Pax3 expression, and no
expres-sion of early determinants of Schwann cells and satellite glia in
mouse DRG and peripheral nerves. In contrast, other
neural-crest-derived lineages, such as sensory neurons and those in the cardiac
neural crest, although reduced in number, were specified in DRG
and the branchial arches and heart outflow tract, respectively. This
finding is consistent with the study of
Ignatius et al. (2013)
showing
that DRG neurons are specified in HDAC1 mutant zebrafish, but are
reduced in number, particularly in the tail, and appears in partial
contrast to the absence of differentiation of cardiac neural crest in the
branchial arches of HDAC1 mutant zebrafish (Ignatius et al., 2013).
To analyze the molecular mechanisms of HDAC1/2 functions in
the specification of peripheral glia, we used the JoMa1 neural crest
cell line that can be differentiated into glia. In contrast to postnatal
Schwann cells, HDAC1/2 were not able to directly activate Sox10
transcription in JoMa1 neural crest cells. Instead, we identified Pax3
as a critical target gene for HDAC1/2-dependent control of
periph-eral glia specification. Indeed, HDAC2 and Sox10 were bound to and
activated the Pax3 promoter in JoMa1 neural crest cells. Of note, the
1.6 kb Pax3 promoter contains two known neural crest enhancer
(NCE) regions (Milewski et al., 2004). However, there are many
putative binding sites for Sox10 in the Pax3 promoter. They are
located in the NCEs, but also outside of the NCE regions. For this
reason, we did not analyze the binding of HDAC2/Sox10 complexes
to the NCEs in particular, but to the Pax3 promoter in general.
Our study shows that Sox10 recruits HDAC1/2 to the Pax3
and P0 promoters in neural crest cells to induce their
transcrip-tion and that HDAC1/2 are essential for expression of these target
genes. In turn, Pax3 binds to and activates the Sox10 MCS4
en-hancer that is critical for Sox10 expression in neural crest cells
during specification of the glial lineage (Wahlbuhl et al., 2012).
Pax3 is necessary to maintain high Sox10 levels and therefore to
trigger appropriate expression of the early determinants of
Schwann cells and satellite glia. In the absence of HDAC1/2 in
neural crest cells, Sox10 is strongly reduced, but still expressed in
DRG, indicating that expression of Sox10 in neural crest cells is
not only dependent on HDAC1/2 and Pax3, but also on other
factors, as suggested by the multiple binding sites for several
tran-scription factors located on Sox10 MCS4 (Werner et al., 2007;
Wahlbuhl et al., 2012). However, we demonstrate that HDAC1/
2-dependent control of Sox10 MCS4 activity is mediated by Pax3.
In addition to controlling Pax3 expression, we found that
HDAC1/2 also control P0 expression directly. Therefore,
HDAC1/2 induce the specification of neural crest cells into
pe-ripheral glia by controlling the expression of the transcription
factors necessary for this process, but also the expression of some
of their key target genes.
We show that the Sox10 MCS4 enhancer is active in JoMa1
neural crest cells, whereas the Sox10 promoter is not. In contrast,
the Sox10 promoter is active in postnatal Schwann cells, whereas
the Sox10 MCS4 enhancer is not. These different activities of gene
regulatory regions depending on the developmental stage are
likely due to the availability and/or recruitment of different
co-factors at these distinct stages. Different Sox10-HDAC1/2
com-plexes must thus be formed and recruited to distinct gene
regulatory regions. Although challenging, it will be very
interest-ing to identify other potential bindinterest-ing partners of the distinct
Sox10-HDAC1/2 complexes at different developmental stages of
Figure 8. In the absence of HDAC1/2, sensory neurons are specified and cardiac neural crest
cells express differentiation markers.A, Confocal pictures of NeuroD1 (red) and Islet1 (green) coimmunofluorescence in E10 and E10.5 control and HDAC1/2 dKO DRG. Overlay of NeuroD1 and Islet1 appears yellow.B, Confocal pictures of Islet1 (green) and cCasp3 (red) coimmunofluores-cence and DAPI (blue) labeling in E10.5 control and HDAC1/2 dKO DRG. White arrows indicate cCasp3-positive DRG cells.C, In vivo cell fate mapping by X-Gal staining (black arrowheads) using the recombination-dependent Rosa26-lacZ reporter allele, Sox9 in situ hybridization (black arrowheads), and SMA (green, open arrowheads), cCasp3 (red, white arrowheads) co-immunofluorescence, and DAPI (blue) labeling in E10.5 control and HDAC1/2 dKO heart outflow tract. The white arrow shows nonspecific auto-fluorescence signal. Scale bar, 100m.Foreach panel, representative photographs of three controls and three dKOs for each developmental stage are shown.
the Schwann cell lineage that may be responsible for the
recruit-ment of these complexes to different gene regulatory regions.
In summary, we demonstrate that HDAC1/2 are required for
Pax3 expression in neural crest cells and that Pax3 is necessary for
HDAC2-dependent control of Sox10 levels and expression of
early determinants of Schwann cells and satellite glia. In addition,
HDAC1/2 control the expression of P0 directly. By regulating the
interplay between Sox10 and Pax3 and the expression of their
target genes, HDAC1/2 are thus essential for specification of
neu-ral crest cells into peripheneu-ral glia.
References
Antonellis A, Huynh JL, Lee-Lin SQ, Vinton RM, Renaud G, Loftus SK, Elliot G, Wolfsberg TG, Green ED, McCallion AS, Pavan WJ (2008) Identifi-cation of neural crest and glial enhancers at the mouse Sox10 locus through transgenesis in zebrafish. PLoS Genet 4:e1000174.CrossRef Medline
Auerbach R (1954) Analysis of the developmental effects of a lethal mutation in the house mouse. Journal of Experimental Zoology 127:305–329.CrossRef Britsch S, Goerich DE, Riethmacher D, Peirano RI, Rossner M, Nave KA,
Birchmeier C, Wegner M (2001) The transcription factor Sox10 is a key regulator of peripheral glial development. Genes Dev 15:66 –78.CrossRef Medline
Campeau E, Ruhl VE, Rodier F, Smith CL, Rahmberg BL, Fuss JO, Campisi J, Yaswen P, Cooper PK, Kaufman PD (2009) A versatile viral system for expression and depletion of proteins in mammalian cells. PLoS One 4:e6529.CrossRef Medline
Danielian PS, Muccino D, Rowitch DH, Michael SK, McMahon AP (1998) Modi-fication of gene activity in mouse embryos in utero by a tamoxifen-inducible form of Cre recombinase. Curr Biol 8:1323–1326.CrossRef Medline
de Ruijter AJ, van Gennip AH, Caron HN, Kemp S, van Kuilenburg AB (2003) Histone deacetylases (HDACs): characterization of the classical HDAC family. Biochem J 370:737–749.CrossRef Medline
Doddrell RD, Dun XP, Moate RM, Jessen KR, Mirsky R, Parkinson DB (2012) Regulation of Schwann cell differentiation and proliferation by the Pax-3 transcription factor. Glia 60:1269 –1278.CrossRef Medline Epstein JA, Lam P, Jepeal L, Maas RL, Shapiro DN (1995) Pax3 inhibits
myogenic differentiation of cultured myoblast cells. J Biol Chem 270: 11719 –11722.Medline
Franz T (1990) Defective ensheathment of motoric nerves in the Splotch mutant mouse. Acta Anat 138:246 –253.CrossRef Medline
Glozak MA, Sengupta N, Zhang X, Seto E (2005) Acetylation and deacety-lation of non-histone proteins. Gene 363:15–23.CrossRef Medline Greenwood AL, Turner EE, Anderson DJ (1999) Identification of dividing,
determined sensory neuron precursors in the mammalian neural crest. Development 126:3545–3559.Medline
Hagedorn L, Suter U, Sommer L (1999) P0 and PMP22 mark a multipotent neural crest-derived cell type that displays community effects in response to TGF-beta family factors. Development 126:3781–3794.Medline Hari L, Miescher I, Shakhova O, Suter U, Chin L, Taketo M, Richardson WD,
Kessaris N, Sommer L (2012) Temporal control of neural crest lineage generation by Wnt/-catenin signaling. Development 139:2107–2117. CrossRef Medline
Ignatius MS, Unal Eroglu A, Malireddy S, Gallagher G, Nambiar RM, Henion PD (2013) Distinct functional and temporal requirements for zebrafish Hdac1 during neural crest-derived craniofacial and peripheral neuron development. PLoS One 8:e63218.CrossRef Medline
Jacob C, Christen CN, Pereira JA, Somandin C, Baggiolini A, Lo¨tscher P, Ozc¸elik M, Tricaud N, Meijer D, Yamaguchi T, Matthias P, Suter U (2011) HDAC1 and HDAC2 control the transcriptional program of my-elination and the survival of Schwann cells. Nat Neurosci 14:429 – 436. CrossRef Medline
Jessen KR, Mirsky R (2005) The origin and development of glial cells in peripheral nerves. Nat Rev Neurosci 6:671– 682.CrossRef Medline Kioussi C, Gross MK, Gruss P (1995) Pax3: a paired domain gene as a
reg-ulator in PNS myelination. Neuron 15:553–562.CrossRef Medline Kuhlbrodt K, Herbarth B, Sock E, Hermans-Borgmeyer I, Wegner M (1998)
Sox10, a novel transcriptional modulator in glial cells. J Neurosci 18:237– 250.Medline
Kurtz A, Zimmer A, Schnu¨tgen F, Bru¨ning G, Spener F, Mu¨ller T (1994) The
expression pattern of a novel gene encoding brain-fatty acid binding pro-tein correlates with neuronal and glial cell development. Development 120:2637–2649.Medline
Lang D, Epstein JA (2003) Sox10 and Pax3 physically interact to mediate activation of a conserved c-RET enhancer. Hum Mol Genet 12:937–945. CrossRef Medline
Ma Q, Fode C, Guillemot F, Anderson DJ (1999) Neurogenin1 and neuro-genin2 control two distinct waves of neurogenesis in developing dorsal root ganglia. Genes Dev 13:1717–1728.CrossRef Medline
Maurer J, Fuchs S, Ja¨ger R, Kurz B, Sommer L, Schorle H (2007) Establish-ment and controlled differentiation of neural crest stem cell lines using conditional transgenesis. Differentiation 75:580 –591.CrossRef Medline Michan S, Sinclair D (2007) Sirtuins in mammals: insights into their
biolog-ical function. Biochem J 404:1–13.CrossRef Medline
Milewski RC, Chi NC, Li J, Brown C, Lu MM, Epstein JA (2004) Identifica-tion of minimal enhancer elements sufficient for Pax3 expression in neu-ral crest and implication of Tead2 as a regulator of Pax3. Development 131:829 – 837.CrossRef Medline
Olaopa M, Zhou HM, Snider P, Wang J, Schwartz RJ, Moon AM, Conway SJ (2011) Pax3 is essential for normal cardiac neural crest morphogenesis but is not required during migration nor outflow tract septation. Dev Biol 356:308 –322.CrossRef Medline
Paratore C, Suter U, Sommer L (1999) Embryonic gene expression resolved at the cellular level by fluorescence in situ hybridization. Histochem Cell Biol 111:435– 443.CrossRef Medline
Paratore C, Goerich DE, Suter U, Wegner M, Sommer L (2001) Survival and glial fate acquisition of neural crest cells are regulated by an interplay between the transcription factor Sox10 and extrinsic combinatorial sig-naling. Development 128:3949 –3961.Medline
Peirano RI, Wegner M (2000) The glial transcription factor Sox10 binds to DNA both as monomer and dimer with different functional conse-quences. Nucleic Acids Res 28:3047–3055.CrossRef Medline
Peirano RI, Goerich DE, Riethmacher D, Wegner M (2000) Protein zero gene expression is regulated by the glial transcription factor Sox10. Mol Cell Biol 20:3198 –3209.CrossRef Medline
Prasad MK, Reed X, Gorkin DU, Cronin JC, McAdow AR, Chain K, Hodonsky CJ, Jones EA, Svaren J, Antonellis A, Johnson SL, Loftus SK, Pavan WJ, McCallion AS (2011) SOX10 directly modulates ERBB3 transcription via an intronic neural crest enhancer. BMC Dev Biol 11:40.CrossRef Medline
Sanchez-Ferras O, Coutaud B, Djavanbakht Samani T, Tremblay I, Souch-kova O, Pilon N (2012) Caudal-related homeobox (Cdx)-dependent integration of canonical Wingless/int1-related (Wnt) signals on a well-conserved neural crest enhancer of the proximal Paired-box 3 (Pax3) promoter. J Biol Chem 287:16623–16635.CrossRef Medline Shah NM, Marchionni MA, Isaacs I, Stroobant P, Anderson DJ (1994) Glial
growth factor restricts mammalian neural crest stem cells to a glial fate. Cell 77:349 –360.CrossRef Medline
Soriano P (1999) Generalized lacZ expression with the ROSA26 Cre re-porter strain. Nat Genet 21:70 –71.CrossRef Medline
Stemple DL, Anderson DJ (1992) Isolation of a stem cell for neurons and glia from the mammalian neural crest. Cell 71:973–985.CrossRef Medline Wahlbuhl M, Reiprich S, Vogl MR, Bo¨sl MR, Wegner M (2012) Transcrip-tion factor Sox10 orchestrates activity of a neural crest-specific enhancer in the vicinity of its gene. Nucleic Acids Res 40:88 –101.CrossRef Medline Wang Z, Zang C, Cui K, Schones DE, Barski A, Peng W, Zhao K (2009) Genome-wide mapping of HATs and HDACs reveals distinct function in active and inactive genes. Cell 138:1019 –1031.CrossRef Medline Werner T, Hammer A, Wahlbuhl M, Bo¨sl MR, Wegner M (2007) Multiple
conserved regulatory elements with overlapping functions determine Sox10 expression in mouse embryogenesis. Nucleic Acids Res 35:6526 – 6538.CrossRef Medline
Yamaguchi T, Cubizolles F, Zhang Y, Reichert N, Kohler H, Seiser C, Matthias P (2010) Histone deacetylases 1 and 2 act in concert to promote the G1-to-S progression. Genes Dev 24:455– 469.CrossRef Medline Yao YL, Yang WM (2011) Beyond histone and deacetylase: an overview of
cytoplasmic histone deacetylases and their nonhistone substrates. J Biomed Biotechnol 2011:146493.CrossRef Medline
Zhang J, Xu F, Hashimshony T, Keshet I, Cedar H (2002) Establishment of transcriptional competence in early and late S phase. Nature 420:198 – 202.CrossRef Medline
